Anyone got good test cases for legacy inputs and case changing? github.com/latex3/latex2e/issues/881

@DavidCarlisle ouch! I'll look into it ASAP! Thx!

@JosephWright No test case, but I wish you luck in fixing the issue!

@mickep :)

@mickep I'm pretty confident what I have is right ...

@JosephWright Very good. :)

@DavidCarlisle I've checked in the expl3 side of the plan, so we can test that it's safe (enough)

@DavidCarlisle Of course, I found that I'd forgotten to update the LaTeX2e version in the l3 repo ...

@DavidCarlisle the chess example here davidcarlisle.github.io/latexcgi/test2-make4ht is that using some local xskak.4ht file? I can't compile it on my system.

@JosephWright @UlrikeFischer making test files is a really bad idea

@DavidCarlisle which one failed?

@UlrikeFischer yes github.com/davidcarlisle/latexcgi/tree/main/make4ht

@UlrikeFischer one I just wrote

@DavidCarlisle ???

@UlrikeFischer all the 4ht files there provided by michal

@JosephWright without running it, what is the output of test3 here?

\documentclass{article}

%\input{test2e}%\errorstopmode

\usepackage[T2A]{fontenc}

\def\TEST#1{\sbox0{#1}\showbox0 \usebox{0}}

\begin{document}

\TEST{0: \CYRA\CYRB\CYRV}

\TEST{1: \MakeUppercase{\cyra\cyrb\cyrv}}

\ExplSyntaxOn
\TEST{2:~ \text_uppercase:n{\cyra\cyrb\cyrv}}
\ExplSyntaxOff

\newcommand\foo{abc\cyra\cyrb\cyrv}

\TEST{3: \MakeUppercase{lower: \protect\foo\ upper: \foo}}
\end{document}

@DavidCarlisle I assume lower: abc[\cyra\cyrb\cyrv] upper: abc[\CYRA\CYRB\CYRV]

@JosephWright boing, you lose:-)

@DavidCarlisle Not entirely surprised: the input is poorly set up :)

@DavidCarlisle I guess I'm missing a path for \protect: as we are updating anyway ...

@DavidCarlisle It's expanding correctly, I must be missing something in the case changer

@JosephWright My main complaint is with an old latex, I don't think the output is jusifiable: the local LICR redfinitions are in force during the edef (good, so \protect\foo not upprcased then) but also still in force when the internal macro executed so abc is lowrcase still but cyra\cyrb\cyrv is uppercase

@JosephWright I'll add a \protect test to the pure l3 call....

@JosephWright your l3 version gets it right even though it is a chang in behaviour

\documentclass{article}

\scrollmode
%\input{test2e}%\errorstopmode

\usepackage[T2A]{fontenc}

\def\TEST#1{\sbox0{#1}\showbox0 \usebox{0}}

\begin{document}

\TEST{0: \CYRA\CYRB\CYRV}

\TEST{1: \MakeUppercase{\cyra\cyrb\cyrv}}

\ExplSyntaxOn
\TEST{2:~ \text_uppercase:n{\cyra\cyrb\cyrv}}
\ExplSyntaxOff

\newcommand\foo{abc\cyra\cyrb\cyrv}

\TEST{3: \MakeUppercase{lower: \protect\foo\ upper: \foo}}


\ExplSyntaxOn
\TEST{4:~ \text_uppercase:n{lower: \protect\foo\ upper: \foo}}
\ExplSyntaxOff
\end{document}

@DavidCarlisle 'Fixes'

@DavidCarlisle Ah, yes: I see what's going on - it's the extra \protected@edef

@JosephWright yes:-) Better test can you make all these pass?

\documentclass{article}

\setlength\parindent{0pt}
\setlength\parskip{20pt}

\ExplSyntaxOn
\let\textuppercase:\text_uppercase:n
\ExplSyntaxOff

\scrollmode
%\input{test2e}%\errorstopmode

\usepackage[T2A]{fontenc}

\def\TEST#1#2{\sbox0{#1}\showbox0 \usebox{0}\\$\rightarrow$ #2}

\begin{document}

\TEST{0: \CYRA\CYRB\CYRV}
         {\CYRA\CYRB\CYRV}

\TEST{1: \MakeUppercase{\cyra\cyrb\cyrv}}
         {\CYRA\CYRB\CYRV}

\TEST{2: \textuppercase{\cyra\cyrb\cyrv}}
         {\CYRA\CYRB\CYRV}

\newcommand\foo{abc\cyra\cyrb\cyrv}

spurious : in the \let

oh sorry it does already pass, if we just use the l3 version without the edef (2 and 4 here)

@DavidCarlisle Yup, looks good to me too

@JosephWright new branch/pr and we discard current one, or iterate in the current feature branch?

@DavidCarlisle As we can squash down anyway, probably just modify current one - it will need updated expl3 I guess for legacy encodings, so I'd appreciate a check at the expl3 end so I can get that out

@JosephWright make latex3 branch with the update and I'll try some pure l3 tests later in that branch?,I need to drop out for a bit

@DavidCarlisle I'll simply do an update of dev to the burning-edge expl3 - we are definitely going this way

@DavidCarlisle I'll sort it shortly

@JosephWright OK

ooh more broken things

Not all hope for a big PL100 party is lost :)

@samcarter yay, gotta get myself a party sombrero :)

@PauloCereda Oh no.

@Skillmon ooh

@PauloCereda counter-ooh

@Skillmon <3

@Skillmon \hat[sombrero]{Skillmon}

@samcarter Don't click here

@PauloCereda :P

ooh tacos

@PauloCereda counter-ooh!

@PauloCereda ooh, a carrot! Here's one for you as well: <3

@Skillmon ooh

ooh a carrot showing in the other direction: ε>

@Skillmon in the spirit of sombreros, check Signal :)

@DavidCarlisle Should be good now :)

@yo' thanks

@JosephWright Just getting back to tex... I got a bit lost what's the current state?

@DavidCarlisle I've pushed the expl3 change, nothing else

@JosephWright in latex3 main branch? so I could do a local install from there?

@DavidCarlisle yes

@JosephWright I got sidetracked into re-activating the openmath.org website this afternoon, complicated travis scripts syncing three different repos have surprisingly stopped working since anyone last checked the back end 4 years ago....

Running l3build with target "install" for module l3cctab
Error: Cannot find configuration build.lua

@JosephWright l3build install in latex3/latex3 (works fine in l3kernel subdir, which is OK for now)

@DavidCarlisle Old directory you need to remove

@JosephWright ah git failure on my part I believed git status not complaining:-)

@DavidCarlisle Git tracks files not folders, but for ease we just use a directory listing for the bundles ....

@JosephWright yes I know that in theory, just not when looking at l3build log errors

looking good so far, making a test with latin 1 input and new format working better

@DavidCarlisle what is now the new definition of \MakeUppercase?

Ohh, trick questions. :)

@UlrikeFischer not done yet but \let\MakeUppercase\text_uppercase:n

@DavidCarlisle Well probably still need \DeclareRobustCommand, but yes

@JosephWright not \protected\def ? :-)

@DavidCarlisle Well yes ...

@JosephWright do you think one can handle this somehow: tex.stackexchange.com/a/523707/2388 ? It seems to need the language

@UlrikeFischer That's where the BCP47 plan comes in: I am hoping for \text_<thing>case:Vn \WhateverPlacebabelUses {#1} longer-term

@UlrikeFischer I'd wondered about offering \MakeUppercase*{<BCP47>}{<string>}

@JosephWright perhaps as optional argument \MakeUppercase*[bcp]{string}, then it could retrieve the default bcp from babel. (which reminds me, there is an open discussion about that isn't there?)

@JosephWright or ` \MakeUppercase[<BCP47>]{<string>}`?

@DavidCarlisle Yes, or that

@UlrikeFischer Yes, that would work too: we got a plan for babel, but I got nothing so far from polyglossia

@DavidCarlisle, @UlrikeFischer I think the plan is clear, I was hoping to get this sorted out for the Fall release - sound OK?

@JosephWright you mean the bcp stuff? Sure.

@UlrikeFischer Cool - I guess we could have done that for the current round, but there was quite a high tariff to the changes we did make

menukeys' documentation no longer builds :( And I'm afraid I can't even complain

@JosephWright well, it uses l3doc.ist...

@Skillmon hm \, in the one donotindex looks fishy. Did that really work?

@UlrikeFischer cause not, locally there is a comma behind it (though I'm not sure for what it was added, could be I guessed wrong and it should be `\`)

@Skillmon That's not changed recently ...

@JosephWright stuff works with 2021's TL. Issue is related to the hyperref changes to doc I guess.

@Skillmon Probably: so blame FMi

@JosephWright well, I get this:

[36] (./menukeys.gls) [37] (./menukeys.ind
! Use of \hyper@link@ doesn't match its definition.
\@ifnextchar ... \reserved@d =#1\def \reserved@a {
                                                  #2}\def \reserved@b {#3}\f...
l.19 ...\arrowkey+\pfill \hdclindex{46}{usage}{13}

@Skillmon Not case changing so Not My Fault

@JosephWright not the PR I merged, so not mine either :) Maybe I'll tackle this tomorrow. It's bedtime here :)

Unfortunately it's also not l3doc.ist, get the same error using gind.ist.

@Skillmon Night